Improving the Precision of a Text Retrieval System with Compound Analysis

In this paper we describe research on compound analysis in the UPLIFT information retrieval project. Results of earlier experiments indicated that splitting up compounds in the query and forming new compounds by combining query terms improves recall while precision does not deteriorate. We investigated whether adding syntactic constraints to the compound splitting and formation processes would improve our initial results. We compared different strategies for compound formation and we also investigated the e ect of adding compound constituents as separate